Techniques for Deriving And/Or Leveraging Application-Centric Model Metric

2. The method of claim 1, wherein the linking and the identifying are practiced in connection with the representations of the performance manifolds of the respective prediction models.

3. The method of claim 1, further comprising, for each of the different prediction models, generating prototype exemplars for each of the created sub-models, the prototype exemplars for each created sub-model being objects to which the respective model can be applied to result in a match with the respective sub-model, the prototype exemplars characterizing the volume and/or shape for an estimated portion of the performance manifold.

4. The method of claim 3, wherein the objects are images and/or image collections.

5. The method of claim 1, wherein the generating of the representations is performed prior to reception of the input.

6. The method of claim 1, further comprising determining which features are strongly correlated with performance of the model by receiving a user-specified list of one or more features and/or by running a residual network feature extractor.

7. The method of claim 1, wherein the training data sets and/or the input include(s) geospatial and/or geotemporal data.

8. The method of claim 1, wherein each of the different prediction models is trained based on a different training data set.

9. The method of claim 1, wherein the indication of the region is defined as a set of one or more attributes that parameterize at least one of the different models.

10. The method of claim 1, wherein the indication of the region is defined using a plurality of images.

11. The method of claim 1, wherein the features parameterizing the performance manifolds include spatial extent, National Imagery Interpretability Rating Scale (NIIRS), off-nadir angle, signal-to-noise ratio (SNR), and/or cloud coverage values.

12. The method of claim 1, wherein the expected performance of the models reflects accuracy of identifying an object of a given type from new and/or unseen images.

14. The non-transitory computer readable storage medium of claim 13, wherein the training data sets used to train the different prediction models are representable as a set of first locations in the respective performance manifolds.

15. The non-transitory computer readable storage medium of claim 14, wherein further data that is not used to train the different prediction models is representable as a set of second locations in the respective performance manifolds.

16. The non-transitory computer readable storage medium of claim 13, wherein for each of the different prediction models, prototype exemplars are generated for each of the created sub-models, the prototype exemplars for each created sub-model being objects to which the respective model can be applied to result in a match with the respective sub-model, the prototype exemplars characterizing the volume and/or shape for an estimated portion of the performance manifold.

17. The non-transitory computer readable storage medium of claim 13, wherein the expected performance of the models reflects accuracy of identifying an object of a given type from new and/or unseen images.

19. The system of claim 18, wherein the linking and the identifying are practiced in connection with the representations of the performance manifolds of the respective prediction models.

20. The system of claim 18, wherein for each of the different prediction models, prototype exemplars are generated for each of the created sub-models, the prototype exemplars for each created sub-model being objects to which the respective model can be applied to result in a match with the respective sub-model, the prototype exemplars characterizing the volume and/or shape for an estimated portion of the performance manifold.

21. The system of claim 20, wherein the objects are images and/or image collections, the objects being parameterized explicitly on the features.

22. The system of claim 18, wherein the generating of the representations is performed prior to reception of the input.

23. The system of claim 18, wherein the determination as to which features are strongly correlated with performance of the model is made based on a user-specified list of one or more features and/or by running a residual network feature extractor.

24. The system of claim 18, wherein the training data sets and/or the input include(s) geospatial and/or geotemporal data.

25. The system of claim 18, wherein each of the different prediction models is trained based on a different training data set.

26. The system of claim 18, wherein the indication of the region is defined as a set of one or more attributes that parameterize at least one of the different models.

27. The system of claim 18, wherein the indication of the region is defined using a plurality of images.

28. The system of claim 18, wherein the features parameterizing the performance manifolds include spatial extent, National Imagery Interpretability Rating Scale (NIIRS), off-nadir angle, signal-to-noise ratio (SNR), and/or cloud coverage values.

29. The system of claim 18, wherein the expected performance of the models reflects accuracy of identifying an object of a given type from new and/or unseen images.